Cross-border payouts at agent speed

Sending money across a border is still slow and expensive for reasons that have little to do with technology. It passes through correspondent banks, each with its own cut-off times, compliance checks and fees. It converts currencies at rates the sender rarely sees. It arrives in a day, or three, or after someone chases it. People have learned to tolerate this because they send money across borders occasionally, and a person can wait.

Software cannot, or rather should not. When agents pay suppliers, contractors and services across borders, they do it often, in small amounts and as part of work that is otherwise finished in seconds. At that frequency, a cost that a person shrugs off becomes a structural tax, and a delay becomes a stalled process. The argument here is that this shifts cross-border payments from a policy problem, addressed slowly through coordination between central banks, to an engineering problem that can be solved route by route, today, inside the regulated system.

The targets that will be missed

In October 2021, the Financial Stability Board set global targets for the G20 roadmap, most of them for end-2027 (FSB). For retail payments, including business-to-business: a global average cost of no more than 1%, with no corridor above 3%, and 75% of payments making funds available to the recipient within one hour of initiation, with the rest within one business day. For remittances: a global average cost of sending USD 200 of no more than 3% by 2030, reaffirming the UN Sustainable Development Goal, with no corridor above 5%, and the same one-hour speed target by end-2027. All payment service providers should disclose total cost, expected delivery time, payment status tracking and terms of service.

The FSB's 2025 progress report, published on 9 October 2025, is frank. Despite significant progress, efforts "have not yet translated into tangible improvements for end-users at the global level", and it is "unlikely that satisfactory improvements at the global level will be achieved in line with the 2027 Roadmap timetable." The speed of wholesale payments and of remittances has improved. The average cost of remittances "remains sticky".

The World Bank's measure shows why. The global average cost of sending USD 200 rose from 6.26% in the fourth quarter of 2024 to 6.49% in the first quarter of 2025 (World Bank), more than double the 3% goal.

A new rail has arrived, unevenly

Stablecoins are now large enough to matter and regulated enough to use. In the United States, the President signed the GENIUS Act, S. 1582, into law on 18 July 2025, creating a federal framework for payment stablecoins (White House). In Hong Kong, the Stablecoins Ordinance came into operation on 1 August 2025, with a licensing regime run by the Hong Kong Monetary Authority (Morgan Lewis).

Measuring how much they are actually used for payments is harder than the headlines suggest. Visa's analysis shows how much raw on-chain volume is noise: over one 30-day period, USD 3.9 trillion of stablecoin transaction volume fell to USD 817.5 billion once bot activity and inorganic flows were removed, and retail-sized transactions were less than 1% of adjusted volume in the twelve months to March 2025 (Visa). The rail is big, but most of what moves on it is not yet everyday commerce.

The sceptical case deserves a hearing too. The Bank for International Settlements argued in its 2025 Annual Economic Report that stablecoins perform poorly on singleness, elasticity and integrity, and "may at best serve a subsidiary role" in the monetary system, while acknowledging their use for cross-border payments, particularly by residents of emerging markets without access to dollar accounts.

Both views can be true. Stablecoins need not replace the monetary system to be the fastest leg of a particular payout.

Why it becomes an engineering problem

Policy improves the average. An engineer only has to improve the route in front of them.

A cross-border payout is a chain of steps: collect funds in one currency, convert, move value across the border, convert again if needed, and deliver into an account the recipient can use. Each step has options with different costs, speeds and cut-off times. A bank wire bundles them into one opaque product. An orchestration layer can unbundle them: receive into a local virtual account, convert to a stablecoin where that is cheaper and permitted, settle across the border in minutes, and pay out locally, with each regulated act performed by a licensed provider.

For a person sending money twice a year, that optimisation is not worth the effort. For an agent paying hundreds of small invoices a week across a dozen countries, it is the whole difference between a process that works and one that waits on correspondent banks. And it can be done per payment, choosing the cheapest compliant route at the moment of sending, which no policy roadmap can do.

What this means for executives

If your business pays people or suppliers abroad, treat the route as a decision rather than a default. Ask for the total cost of each payout, including foreign exchange, not just the visible fee, and for delivery time and status tracking: the same disclosures the FSB expects of every provider by end-2027. Where agents will pay on your behalf, make sure the payment layer exposes those numbers to software, so the agent can choose the route and you can see what it chose.

What this means for policymakers

The 2027 targets are about outcomes for end users, and outcomes can now improve faster in the private orchestration layer than through infrastructure reform alone. Policy can help by keeping the regulated perimeter clear: licensed providers performing conversion and payout, clear rules for stablecoin issuers, and disclosure rules that apply whether the caller is a person or software. Measurement matters as well. The FSB notes that data on transparency remains insufficient. Requiring providers to report realised cost and speed per corridor, including for programmatic payouts, would show where the new rails are closing the gap and where they are not.

Where Praecise fits

Skerry provides payment rails for agents, including stablecoin settlement. It is global payment orchestration: one programmable surface over virtual accounts, stablecoin conversion and cross-border payouts. Bridge and Stripe perform the regulated acts under their own licences. With Ipnops and Praecise Inbox, businesses hand payments, books, orders and customer answers to agents across the accounts they already use, and teams building agents on Praecise Cloud give each agent an inbox and a way to pay.